Assessment of diagnostic accuracy and inter-pathologist agreement in telecytology performed for canine round cell tumours using images captured by a smartphone camera.

Abstract

Neoplasia is a common disease in companion animals. Rapid and accurate diagnosis of neoplasia often leads to a better disease outcome. Cytology is an inexpensive and minimally invasive method frequently used to diagnose neoplasia. However, the physical distance between veterinary practices and pathologists often impedes rapid diagnosis. Telecytology facilitates rapid interpretation of cytology results leading to prompt diagnosis. Smartphone cameras have been suggested as an alternative to image-capturing devices used in conventional telecytology. This project was designed to evaluate whether telecytology with smartphone cameras can be effectively used to diagnose canine round cell neoplasms. For this purpose, five round cell tumours [lymphoma, mast cell tumour (MCT), transmissible venereal tumour (TVT), plasmacytoma and histiocytoma] were imaged using a smartphone camera and the images were sent to two pathologists at distant locations via emails. For each tumour type, the cytopathological diagnosis from the two pathologists was compared with the histopathological diagnosis of the same tumour and the agreement was determined by Cohen's kappa test. Inter-pathologist variability was also determined using the same test. The diagnostic agreement between histopathology and cytopathology was excellent (κ = 1.0) for lymphoma and MCT when x10 objective images were used. Variable agreements were reported for TVT, plasmacytoma and histiocytoma. When both 10x and 40x objective images were used, diagnostic agreement between histopathology and cytopathology was excellent for all tumours except histiocytoma. The inter-pathologist agreement was excellent for lymphoma, MCT and TVT using the 10x objective, while with the 40x objective it was excellent for all tumours except histiocytoma. The diagnostic accuracy was greater when both 10x and 40x objective images were used for diagnosis than when only 10x objective images were used (χ2 = 40.0, p = < 0.001). The use of images obtained using 10x and 40x objectives increased the diagnostic accuracy compared to the use of only 10x objective. These findings suggest that telecytology performed with smartphone cameras is a promising method to rapidly diagnose canine round cell tumours. Future studies with larger number of samples and samples with atypical presentations of round cell neoplasms will be helpful to further elaborate the present findings.